Female directors relationship to financial performance. : A study of female directors impact on financial performance and the presence of "glass cliff" in Sweden.

Female directors’

relationship to

financial performance.

MASTERPROJECT

THESIS WITHIN: Accounting track NUMBER OF CREDITS: 30cr MSc

PROGRAMME OF STUDY: Civilekonomprogrammet AUTHOR: Oscar Ahldén and Felix Kollberg

JÖNKÖPING May 2018

A study of female directors impact on financial

i

Master Thesis

Degree Project in Business Administration

Title: Female directors’ relationship to financial performance. Authors: Oscar Ahldén and Felix Kollberg

Date: 2018-05-21

Key terms: Female representation, Gender diversity, performance measurement, “Glass Cliff”

Abstract

Gender diversity within the boardroom is an important theme in the research of corporate governance. The lack of female directors during recent years have raised attention where the prejudice against women have been a central theme. Especially, in a gender egalitarian country as Sweden. The purpose of the study is to examine the relationship between female directors and financial performance, but also the presence of the “glass cliff” theory. The study is based on companies listed on Large Cap in Sweden, where data are collected from annual reports and a database. The collected data are analyzed by several statistical methods. The findings show that female directors do have a positive impact on a company’s financial performance, in terms of accounting-based measurements. However, the market-based measurement does not a provide a significant relationship to female directors, indicating that the Swedish stock market does neither positively or negatively react to a more gender diverse board. As no differences in performance are seen preceding the appointments of females compared to males, no evidence for the “glass cliff” is found. The findings suggest that female directors may enhance the performance of a company and disproves the prejudice against women. Further, the findings indicate that the Swedish boards are becoming more diverse, where precarious situations are not the reason for women to be appointed to the board.

ii

Table of Contents

... 1

1.

Introduction ... 1

2.

Literature Review and Theoretical framework ...6

2.1 Gender Diversity ...6

2.2 Empirical evidence of Gender Diversity and financial performance ...8

2.3 Empirical evidence of the “Glass Cliff “ ... 12

2.4 Performance measurements ... 16

3.

Regulations of the Board of Directors in Sweden ... 17

4.

Methodology and Method ... 19

4.1 Data Collection and sampling ... 19

4.2 Performance measurements ... 22

4.3 Female directors’ relation to financial performance ... 24

4.4 The “Glass cliff” phenomenon ... 25

5.

Empirical results and Analysis ... 28

5.1 Female directors impact on financial performance ... 28

5.2 Statistical data of the “glass cliff” ... 33

5.3 Annual specific data for gender diversity ... 36

5.4 Limitations and future research ... 40

6.

Conclusion ... 42

iii

Tables

Appendix

1

1. Introduction

_____________________________________________________________________________________

The purpose of this chapter is to introduce the problems females face to be appointed to the board of

directors, but also after they have attained these positions. It introduces a broader view of the problems on an EU-level and then narrow it down to Sweden, in order to provide an understanding for the contemporary situation of gender diversity which is the foundation from where a presentation of the purpose is made.

______________________________________________________________________

1.1 Background

The directors of the board are elected by the shareholders as representatives of their interests and investments. The responsibility of the board of directors is to oversee the overall direction of the company and to elect the management team that deals with the day-to-day operations. The board of directors have also the responsibility to monitor the executives and to ensure that the organization operates in the best interest of all stakeholders. The board of directors should be diverse in its composition in order to be able to comprehend and solve all types of possible issues that may arise. Diversity can be explained in various aspects such as experience, nationality, age and gender (Mishra & Juhnjhunwala, 2013).

Gender diversity has led to a policy debate among European countries and some member states have introduced different kinds of gender quotas (Bianco, Ciavarella, & Signoretti, 2015). Some studies are arguing that these gender quotas have a negative impact on companies’ performance (Ahern & Dittmar, 2012; Gregory-Smith, Main, & O'Reilly III, 2013). For example, Adams, Gray and Nowland (2010) discuss that the new gender quotas may raise costs considering the restriction of the possibility to select the best suitable candidate.

In April 2016, women accounted for just 23.3% of board members of the largest publicly listed companies in the EU-countries, which is low considering the legislation made in former years. Thus, in November 2012, the EU Commission proposed legislation with an aim of reaching a 40% objective of the underrepresented sex (females) in the positions of

non-2

executive directors in publicly listed companies, however small and medium companies are excepted. As this is just an objective to strive for and not a legislation with an actual gender quota, the EU Commission instead focuses on ensuring that the topic is on the political agenda. Increasing the awareness is made by exchanging "good practices" among stakeholders, which are developed by collecting and analyzing data from decision-making figures (European Commission, 2017).

The population of Sweden is often proud of being in the top of the gender egalitarian countries in the world. The World Economic Forum publishes a Gender Gap Index every year and Sweden has been amongst the top four, for seven years as of 2015 (World Economic Forum, 2015). Nevertheless, Sweden is usually seen as contradictory with a simultaneous presence of both gender equality and inequality. Even though the Swedish women and men are moving towards a more equal society, with an equally represented Parliament, a welfare state model based on a dual-earner family etc., women and men face different challenges in the labor market and in the workplace (Holgersson & Wahl, 2017).

Sweden has not yet reached the EU-objective of 40% women in the boardrooms of its publicly listed companies (Ekonomifakta, 2016). Regardless of the thoroughgoing legislation concerning gender-discrimination, there has not been any gender quotas implemented to rectify the imbalances of the corporate board of directors. However, it has been discussed in the parliament to introduce a gender quota legislation, but this have received resistance. The right-wing Members of the Parliament argued that the gender quotas would affect the appointment of board of directors, which is a fundamental right of the shareholders (Holgersson & Wahl, 2017).

The only thing the Swedish Corporate Governance Code (the Code) mention regarding gender equality is that the Swedish companies should strive for a gender diverse board of directors (Swedish Corporate Governance Board, 2016), which is vague compared to the legislation in Norway, with a gender quota of 40% (Bianco et al., 2015). This is indeed a soft law regulation and no sanctions are provided by the Code, if a company should deviate from the proposed target of 40% gender diversity (Holgersson & Wahl, 2017).

Today, the issue of gender diversity is of high importance in Sweden and private corporations are aware of this and develops solutions in order to create legitimacy in the eyes of the

3

stakeholders. Hence, several companies have been in favor for both voluntary and legislated measures in order to achieve an increased number of females in the boardroom (Holgersson & Wahl, 2017). Evidence for the importance of this topic can be found by examining the past, where women consisted of only 2% of the board seats in 1993 and in 2013 women took up 23% of the board seats (Statistics Sweden, 2013). Moreover, on the first of January 2018, the Swedish government introduced the Swedish gender equality agency. The agency's goal is to make sure that gender should not affect a person's ability to shape society and his or her life (Jämställdhetsmyndigheten, 2018). Therefore, measures are being taken, in order to direct resources to prohibit the problems constraining Sweden from achieving gender diversity.

1.2 Problem

There is an ongoing concern regarding the composition of boards in different aspects. The directors of the boards are accused of being dominated by elderly men, benefiting a few connected individuals. The individuals appointed to the board of directors are usually seen as professionals and are sitting on multiple boards for various companies. The connection can affect the independence of the board of directors which is an important feature of the position as directors (Singh, Vinnicombe, & Johnson, 2001).

According Mishra and Juhnjhunwala (2013), the most common and important reason for why boards are still male dominated is the prejudice towards women and their ability to perform equally to men in top-level jobs of the corporate world. Therefore, women with similar or higher qualifications compared to men are not considered suitable for higher positions and are instead offered lower-paying jobs.

Males and females have different behavior, and by extending the boardroom with females, it enhances the human capital of the board of directors (Mishra & Juhnjhunwala, 2013). Due to this and as women have been generally excluded from the “old boys club”, their presence in the boardroom can contribute with different professional background, perspectives and problem-solving skills (Adams & Ferreira, 2009). A heterogeneous board, in terms of gender, is a better monitor of the executive management, than a homogenous board dominantly consisting of males (Adams & Funk, 2012). Furthermore, by having visible presence of females within the board, it may enhance the reputation of the company and some major

4

investors, such as institutional investors, that have a preference for investing in companies with gender diverse boards (Kuczynski, 1999).

Previous research has received different results from their investigation of how gender diversity affects a company. Thus, the focus has been on the effects of companies' performance, which has led to mixed results regarding whether or not gender diverse boards leads to a positive impact on companies’ financial performance. According to Carter, Simkins and Simpson (2003), a positive relationship between gender diversity and accounting-based measures of financial performance exists. Erhardt, Werbel and Shrader (2003) also find a positive relationship between board diversity and financial indicators of firm performance. In contradiction to this, Adams and Ferreira (2009) find a negative relationship between gender diversity and market-based performance. However, they also find that female directors have better attendance records compared to male directors and also that male directors' attendance at board meetings increased the more gender-diverse the board of directors are.

Women's participation in the boardrooms is slightly increasing (Statistics Sweden, 2013), but research also highlights the problems women face when attaining the positions. Instead of investigating if women negatively affect the company’s performance, Ryan and Haslam (2005) believed that it is equally plausible that the poor performance of the company could be the reason for the appointment of women to the board of directors. They constructed and called this phenomenon “the glass cliff”. Their research finds evidence for the phenomenon, consequently women are more preferred in the boardroom when there is an increased risk of negative consequences when elected to the leadership role compared to men. Hennessey, MacDonald and Carroll (2014), ask for further research concerning the topic since the “glass cliff” theory is accepted and contemporary relevant, although without sufficient empirical studies supporting the phenomenon. Therefore, there is a research-relevance gap that requires more empirical evidence and this research will help to fill this gap and to contribute to the discussion of the phenomenon “glass cliff”. In addition, no study has been made in Scandinavian countries, where gender diversity is a relevant topic (Velte, 2018).

5

1.3 Purpose

There are two purposes of this study. First, to examine the relationship between the presence of female directors and financial performance of the Swedish Large Cap companies. Second, to seek evidence for the “glass cliff” phenomenon with data retrieved concerning the companies.

6

2.

Literature Review and Theoretical framework

_____________________________________________________________________________________

The purpose of this chapter is to present previous research on gender diversity and its relation to financial performance to provide a deeper understanding. Further explanation of the “glass cliff” and different aspects that affects earlier research will be introduced.

______________________________________________________________________

2.1 Gender Diversity

The female inequality has been a great concern for the European Union, all the way back to the beginnings of the European Community (the EEC) in 1957 and is one of the founding values of EU. In 1957 the EEC constructed the principle of equal pay for equal work which became a provision of the “Treaty of Rome” (Article 119). Once again in 2007, in the “Lisbon Treaty”, gender equality was one of the core values. As one of the founding values of the European Union, the focus has continued through the years and in the end of 2015, companies should implement one of the recommended methods developed by EU in order to reach better diversity between males and females. As of the end of 2015 females were still underrepresented in leadership positions and held 18% of the board seats (European Commission, 2015).

The major global focus on the inequality for females has also shed light on the gender diversity within the boardroom. The decisions made by the board of directors are of major importance for a company and contributes to its success and long-term survival. Additionally, their decisions affect not only the company, but also its shareholders, customers, employees and other stakeholders. In order to achieve decisions that contributes to long-term success, a diverse board is required. Attributes such as knowledge, expertise, skills and perspectives of the associated market are given by the diversified composition of the board of directors. The board diversity refers to the heterogeneous mixture of age, race, experience, networks etc., and foremost gender (Mishra & Juhnjhunwala, 2013).

Several studies support the positive effects of gender diversity in the boardroom. A heterogeneous board is a better monitor of the executive management team, because of the different viewpoints that the diversity of the board brings (Anderson, Reeb, Upadhyay, & Zhao, 2011). This is in line with the findings from Adams and Funk (2012), however they

7

also find evidence for that women who have obtained leadership positions have significantly different attributes compared to the average woman. They suggest that it is possible that the women who obtain these positions want to be more similar to the male directors, in order to reach these positions and make the gender gap in the boardroom smaller. In addition to different viewpoints and monitoring, a heterogeneous board has a better understanding of complex situations in the company’s environment than a homogenous board and improves the independency (Carter et al., 2003). Moreover, since women have been excluded from the “old boys club”, independence and monitoring functions can be further enhanced by including female directors (Adams & Ferreira, 2009).

As men have occupied the majority of the corporate board seats since the origins of limited companies, the more women added to the board, the more diverse abilities it will gain. Men are seen to have an autocratic leadership style which benefits their control of the management and more focused on the short-term performance and to achieve the objectives of the company. On the other hand, women are seen to have a democratic leadership style with a long-term perspective which enables female directors to build positive and long-lasting relationships with all stakeholders in order for the board to manage risk and be more socially responsible (Mishra & Juhnjhunwala, 2013). Obert, Suppiah, Zororo and Desderio (2015) supports this, by finding that a majority of their interviewees considered women to be more risk averse than men in regard to financial decision-making. In addition, women are considered more socially responsible, more objective and takes decisions with ethical values in consideration. Similarities and differences are found in a survey by Adams and Funk (2012) of directors in Sweden, where female directors are considered friendlier and more universally concerned than male directors, who are more power oriented. In contrast to previous research (Obert et al., 2015), they find female directors more risk loving, but also point out that having women on the board does not lead to more risk-averse decisions (Adams & Funk, 2012).

Previous research find evidence that female directors are less likely to have attendance problems than male directors. Moreover, the higher share of female directors on the board leads to improved attendance behavior by men. Overall, better attendance behavior improves their monitoring function, which contributes to less transaction cost between the directors and causes less agency problem between shareholders and managers (Adams & Ferreira, 2009). The agency problem arises between the shareholders (principal), who are represented

8

by the board of directors, and the executive management (agent) when their interests conflict in how the company operates, also known as the separation of ownership and control (Shleifer & Vishny, 1997). On the contrary, Bianco et al. (2015) find in their empirical study of female representation on Italian boards that female directors perform worse than male directors in attendance to board meetings and an even stronger relationship when the female director were independent.

2.2 Empirical evidence of Gender Diversity and financial performance

As several studies point out the positive effects of a heterogeneous board with better monitoring, independence, attendance etc. (Adams & Ferreira, 2009; Anderson et al., 2011; Carter et al., 2003). Nevertheless, several investigations are conducted to examine if female directors have a positive impact on companies’ performance, using different performance measures.

Singh et al. (2001) investigate the companies of FTSE 100 (an index of the 100 largest listed companies on the London Stock Exchange) and identify the number of women sitting on the board of directors for each company between the years 1999-2000. This identification is compared to the profitability and the turnover of the companies, finding a positive relationship for the companies with the highest profit and turnover and the presence of female directors. Similar results are found by Carter et al. (2003), where a significant positive relationship between firm value and female representation on boards are identified. However, they investigate the US Fortune 1000 firms with a sample of 638 companies from 1997 and using Return on Assets (ROA) and Tobin’s Q-test, a measurement of stock valuation in relation to assets, as performance measurements. In addition, they also find that the number of women increases with firm size but decreases as the number of insiders (both executive and non-executive director) increases. Furthermore, Erhardt et al. (2003) find positive indications of the relationship from both correlation and regression analyzes. The study consists of 127 large US companies between the years 1993-1998 using the accounting-based performance measurements ROA and Return on Investment (ROI) which both were positively associated with board diversity.

Adams and Ferriera (2009), investigate 1939 US listed firms for the time period 1996-2003, using Tobin’s Q-test and ROA. However, their results indicate that female directors have a

9

negative effect on companies’ performance and conclude that no evidence for quota-based policies would improve firm performance and must be motivated by other factors.

Haslam, Ryan, Kulich, Trojanowski and Atkins (2010) conduct a study in order to broaden the base of their earlier study (Ryan & Haslam, 2005). In addition, they wanted to continue the research made by Singh et al. (2001) and conducted an archival examination of the FTSE 100 companies for the time period 2001-2005. However, this study includes both accounting-based measurements, ROA and ROE, and a market-accounting-based measurement, Tobin’s Q-test, instead of just measuring turnover and profitability (Singh et al., 2001). They find no relationship between the presence of women or the percentage of women on the boards and either for ROA or ROE. However, when the same test was conducted with Tobin’s Q-test, they found a negative relationship, especially when testing for at least one woman on the board. During the same year a study was conducted by Adams et. al. (2010), which is in line with the studies by Carter et al. (2003) and Erhardt et al. (2003), and find that the appointment of female directors receives positive market reactions. The study analyzes company announcements, annual reports, records of start date etc. from the Australian Stock Exchange (ASX) and uses stock price from January 1, 2004 to June 30, 2006 as measurement. The author further suggests that female directors bring shareholder value and are even more valuable for firms with greater monitoring needs.

A different approach is used by Ahern and Dittmar (2012), with the purpose to investigate the impact of the gender quota that Norway first passed in 2003, by measuring Tobin’s Q-test of 248 Norwegian public-limited firms for the time-period 2001-2009. They find a negative link between female directors and Tobin’s Q-test during this time-period, due to the gender quotas constraining the boards from maximizing the shareholder value. The gender quota led to younger and less experienced members of the board which negatively affects organizational performance. Further, they suggest that gender quotas are not the way to improve the gender diversity within the boardrooms and a company cannot be forced to achieve the optimal gender mix equilibrium. Joecks, Pull and Vetter (2012), use the “Critical Mass Theory” to examine the relationship. The theory stipulates that until a certain point of women is reached, the focus will not be on what type of skills and abilities females bring to the board. By examining 151 listed German companies within the time period 2000-2005 they find a negative relationship between female directors and ROE. However, the negative relationship is only observable until the “Critical mass” of 30% women is reached. After this

10

point, female directors are associated with higher firm performance compared to boards completely consisting of males.

A more extensive study is conducted by Gregory-Smith et al. (2013) of all the UK listed companies on FTSE350 between the years 1996-2011, where they investigate several factors of the female scarcity in the boardroom. For the investigation of the company’s performance, they measure total shareholder return, ROA, ROE and the price-to-book ratio in order to approximate Tobin’s Q-test. No significant link between the measures and gender diversity is found. The authors refer their findings to the result by Ahern and Dittmar (2012), by saying that maybe the proportion of gender diversity within the board is optimal to start with. Their findings indicate that the probability of appointing a woman to the board depends on whether a woman has stepped down from her position in the recent time, which then enhances the probability. Further they suggest that gender diversity should be achieved, with moral values in consideration (Gregory-Smith et al., 2013).

More recent evidence for a positive relationship between Tobin’s Q-test and gender diversity in the boardroom exist. A significant positive effect is presented in a research of Italian firms from 2011 to 2014 by Giordini and Rancati (2017) and they conclude that improved gender diversity will not negatively affect the shareholder’s value. However, they further suggest that companies should not solely strive for increasing the number of women in the boardroom, but instead reach the optimal mix of males and females. In addition, a research based on US companies between the years 2007-2014 find similar results. The positive relationship is also presented by using ROA as accounting-based measurement. Although, the research points out that if companies have high performance instead of low performance, female directors have stronger quantitative impact (Conyon & He, 2017).

In summary, by using a deductive approach the articles examine the relationship between gender diversity and financial performance and contribute with mixed results. Research conducted from the early 2000s receives more positive relationships between the two variables (Carter et al., 2003; Erhardt et al., 2003; Singh et al., 2001). On the contrary, research conducted around 2010 mostly receive a negative relationship (Adams & Ferreira, 2009; Ahern & Dittmar, 2012; Gregory-Smith et al., 2013). More recent research however, do also find a positive relationship between gender diversity and financial performance (Conyon & He, 2017; Giordini & Rancati, 2017). The mixed results may be due to the different countries

11

(board systems may be different), time periods and performance measurements used. Moreover, the overall representation of female directors affects the relationship between the variables, due to studies conducted with a rather high representation compared to studies with a rather low representation. A study conducted with a large variance of female representation is less likely to find a relationship (Joecks et al., 2012). The sample of articles are mainly based on research conducted from US and UK companies, where different regulations of the board of directors have an impact (Giordini & Rancati, 2017; Velte, 2018). Furthermore, most of the studies combines both Tobin’s Q-test and ROA, in order to receive results from both accounting-based measures and market-based measures.

That females may be mere tokens in the boardroom have been investigated as early as 1977 by Kanter (1993), where she argues that tokens are more visible and seen as representatives of their category. The concept is still relevant and is further investigated in the articles concerning performance, where analyzes are conducted to examine whether the female directors have an impact on the organization’s financial performance or not (Adams & Ferreira, 2009; Gregory-Smith et al., 2013; Joecks et al., 2012).

Even though, no gender quota legislation exists in Sweden, the issue of gender diversity within the boardroom is of high importance which cannot be dismissed by the companies anymore. Furthermore, the work for gender equality in terms of power relation between managers and employees have increased within the organizations and therefore the knowledge and awareness in the society has also increased. One can expect the corporate boards to take actions in order to be in line with the norms of the society’s perceptions. A more equal distribution of gender would therefore have a positive impact on the legitimacy of entities in Sweden. However, the corporate boards are increasing the number of female directors, as this issue grows in the Swedish society, the question is whether they are appointed for their diversity in skills, abilities and knowledge or just as mere tokens (Holgersson & Wahl, 2017).

Studies present mixed results concerning mere tokens, but notably their results are received through different methods. Adams and Ferreira (2009) and Jasmin Joecks et al. (2012), both find evidence that female directors add value to the board of directors through diversity of skills and knowledge and suggests that they are not mere tokens. In addition, Adams et al. (2010) find that the appointment of a female director has a significant and positive market

12

reaction, which also suggests that they are not motivated by tokenism. In contradiction to these findings, Ahern and Dittmar (2012) finds that the corporate boards of Norway have females as mere tokens, because of the gender quota legislation. This is due to the limited pool of female directors, whereas the gender quota legislation led to new younger directors with less experience. Gregory-Smith et al. (2013) further find two factors supporting the concept of tokenism. First, they find evidence that boards have an interest in maintaining a representation of females for the visibility, but the boards do not allow equal opportunity for the appointment of all positions of the non-executive directors. Secondly, they find evidence for tokenism due to the increased likelihood that a female is appointed if another female has recently stepped down from the corporate board.

Based on the evidence presented from previous research included in literature, the two following hypotheses are developed:

H1: There is a positive relationship between female directors and accounting-based measurements of financial performance.

H2: There is a positive relationship between female directors and market-based measurements of financial performance.

2.3 Empirical evidence of the “Glass Cliff “

“Glass ceiling” is a metaphor used to explain the invisible barriers women face when trying to reach higher positions within a company. It was first examined in the 1970´s in the US and is the result of prejudice within organizations. The consequence of the “glass ceiling” is that women are placed in positions associated with lower remuneration and less responsibilities (Wirth, 2001). Results from an investigation in France present evidence of women breaking the “glass ceiling (Dang, Nguyen, & Vo, 2014). Considering that women can break the glass ceiling, research has moved towards what happens after the “glass ceiling”, for example examining the “glass cliff” phenomenon (Ryan & Haslam, 2005). Judge (2003) argue that the presence of women in the boardrooms has a negative impact on the performance of UK companies. In response, Ryan and Haslam (2005) argue the other way around, that the weak performance of organizations enables women to attain high

13

positions containing a higher risk of negative consequences. In order to provide evidence for this, they analyze the FTSE 100 companies during 2003 and compare the appointment of women to men in terms of the stock performance of the companies, by creating a matched sample. The sample contains men appointed approximately at the same time as the women and belongs to the same company sector. They find that when the stock market has a downturn, companies assigning a female director also perform poorly in the months preceding the appointment. On the contrary, the companies with male appointments had a stable performance before the appointments. Thus, Ryan and Haslam (2005) call this phenomenon the “glass cliff”.

Ryan and Haslam (2007) further examine the “glass cliff” theory and the type of discrimination that may be the reason for the phenomenon, such as in-group favoritism, sexism, group dynamics and implicit leadership theories. They also find that women tend to be in leadership positions for a shorter period than men, because women feel that they must take the blame for the poor performance or failure. In addition, women may accept precarious positions due to the fact that men reject these, since the positions can damage their professional reputation. Therefore, women seize this opportunity when these positions are offered (Frazier & Hunt, 1998). Ryan and Haslam (2008) find further evidence for the “glass cliff”, although compared to their earlier research (Ryan & Haslam, 2005), by conducting an experimental study. An experimental study is conducted in an environment where the researcher is able to see how certain variables affect each other, and at the same time control the external variables (Sreejesh, Mohapatra, & Anusree, 2014). Their findings suggest that if men and women are equally qualified for a position, women are more likely to attain the position if the company’s performance is in a downturn (Haslam & Ryan, 2008). Adams, Gupta and Leeth (2009)analyze CEO appointments of companies from the US between 1992-2004, by using the market-based performance measurements, market value and earnings per share (EPS), and the accounting-based measurements, ROA and ROE.To be able to compare between gender, two types of matched male samples are conducted. The first sample contains men appointed as CEO of a company with the same Standard Industrial Classification-code (SIC) as the females’ company, which divide the companies into industries. The second sample contains men appointed to CEO in the same firms appointing a female CEO during the time period. The results oppose the “glass cliff” theory because when a female CEO is appointed, the company tends to have performed well preceding the

14

appointment. In addition, before the appointment of men as CEOs, the companies generally performed poorly. However, they conclude that there is an even playing field for men and women. Ryan and Haslam (2009) respond to Adams et al. (2009) by saying that the playing field for men and women is not equal, which becomes clear when one goes further than just analyzing the financial data, but also examine the psychological underpinnings affecting the CEO appointments.They express that the “glass cliff” may not be a universal problem, but still a frequent problem for women. This is highlighted by a contemporary research in Sweden, which is one of the top egalitarian countries in the world, which claims that men and women face different opportunities in the labor market and in the workplace (Holgersson & Wahl, 2017). To shed further light on the “glass cliff”, another study is conducted by Haslam et al. (2010), where their market-based performance measurements provide evidence for the “glass cliff” with a negative relationship. However, evidence is not found for the accounting-based measurements.

Brady, Isaacs, Reeves, Burroway and Reynolds (2011) take a different direction with their research as they investigate the relationship between how crises and turbulence affect the appointments of women to executive positions. They analyze companies from the US and find a positive relationship between turbulence and scandals compared to the appointment of women, which supports the presence of the “glass cliff” phenomenon.

In line with Ryan and Haslam (2005), Hennessey et al. (2014) use a matched sample for their gender comparisons, which is based on the industry of the company and time of the appointment. They find that the stock market performance for companies, before they appoint women is better compared to the companies appointing men. Therefore, they suggest that there is a so called “solid ledge” instead of a “glass cliff” for women. The authors conclude that previous researches are limited both in number of studies and mainly conducted on companies from US and UK. Hence, they conduct their research on companies from Canada during the years 2006 and 2008 to be able to contribute to the discussion of the “glass cliff” existence. They argue that perhaps by only looking at market-based performance measurements, the results may not be robust enough and most importantly, a research-relevance gap exists regarding the phenomenon. With this in mind they ask for further research, both qualitative and quantitative to contribute to the discussion.

15

Similar to previous research (Adams et al., 2009; Haslam et al., 2010), Elsaid and Ursel (2017) test the glass cliff phenomenon for North American firms by analyzing the relationship between CEO appointments and both market returns and ROA. A matched sample for comparisons between men and female is created. The sample is firstly based on the SIC-code, in line with Adams et al. (2009), which divides the companies into industries. Secondly, the company with the most similar size, measured in total assets, are chosen. The study finds evidence for the “glass cliff”with both of these measurements. However, their findings are not completely in line with the “glass cliff” since they also investigate the tenure of CEOs and find that women are more likely to stay longer in their position as CEO compared to men.

By analyzing the historical development of the research on the “glass cliff” theory it is observable, but also mentioned by researchers, that only a few studies are conducted within this field of research and they primarily use a deductive approach (Hennessey et al., 2014; Velte, 2018). The first study focuses on the appointments of women to the board of directors (Ryan & Haslam, 2005) and later researchers examine CEO appointments (Adams et al., 2009; Elsaid & Ursel, 2017). At the starting point, the investigation is only observational where secondary data are collected from large companies’ websites and analyzes are made on their stock market performance (Ryan & Haslam, 2005). However, as time proceeded, evidence for the phenomenon is also found in experimental studies conducted to analyze the psychological effects of the “glass cliff” (Haslam & Ryan, 2008). Apart from market-based measurements, later researches also use accounting-based measurements (Adams et al., 2009). Previous research find evidence both for and against the phenomenon. A more recent study takes a wider approach and analyze the “glass cliff” with regards to the effects of scandals and turbulence (Brady et al., 2011). Observably, most of the previous research are based on companies located in the US and in the UK, hence research of the “glass cliff” based on other countries is necessary (Hennessey et al., 2014; Velte, 2018). On the basis of the previous research the following hypothesis conducted:

H3: The financial performance preceding the appointments of female directors are equal to the financial performance preceding the appointment of male directors, hence the “glass cliff” phenomenon is not present.

16

2.4 Performance measurements

By using market-based measurements, more variables may affect market performance compared to when using accounting-based measurements of performance. Although, if one on the contrary only analyzes accounting-based measurements, the full picture might not be captured, as it excludes organizational, social and psychological processes in investigations (Ryan & Haslam, 2009). Market reactions have a big impact on the measurement and therefore investors emotions and judgement are of importance (Fama , 1991). In addition, accounting-based measurements are backward-looking, self-composed numbers from the company. Whereas, market-based measurements are explained as forward-looking since they take the present into account, but also capture future expectations.

Accounting-based figures are required to follow laws and regulations in order to minimize the risk of illegal activities such as manipulating the numbers (Haslam et al., 2010). However, the regulations do not stop the companies from using earnings management and therefore the assumption that accounting-based measurements are purely objective is inaccurate (Velte, 2018). A change of the regulations that controls the accounting-based figures can make the measurements harder to compare and less reliable in relation to market-based measurements (Ahern & Dittmar, 2012). Hennessey et al. (2014) highlights that earlier research and their own, frequently use measurements of market performance, such as the share price of a company. They mention that using this type of measurements may not be robust enough when investigating the “glass cliff” phenomenon. Additionally, Ryan et al. (2010) suggest that by using the accounting-based measurements (ROA and ROE) their conclusions become more robust.

The previous research distinguishes their performance measurements into market-based and accounting-based when using multiple measurements within the same category and do not distinguish each measurement specifically. In addition, previous research has never received conflicting results within the same category of measurement. This is taken into consideration when developing H1 and H2.

17

3.

Regulations of the Board of Directors in Sweden

_____________________________________________________________________________________

The purpose of this chapter is to provide the reader with legislated regulations of the Swedish Corporate Governance and the criteria of being listed on Nasdaq.

______________________________________________________________________ The Swedish Corporate Governance Code (the Code) targets all companies with shares listed on a Swedish regulated market, which in 2016 were Nasdaq Stockholm and NGM Equity. Today, the current revised Code is the one of 2016 and applicable to all companies listed in Sweden, even though there can be great differences in their characteristics. The code operates on the principle of comply-or-explain, meaning that a company should follow the recommendations of the Code, or state why it has not complied with the Code and the reasons for deviating from it. The company must also state how they intend to do instead of following the Code (Swedish Corporate Governance Board, 2016). Hence, the Code is a so called soft law, in terms of not being legally binding as companies can deviate from it (Inwinkl, Josefsson, & Wallman, 2015). Contradictory to the soft law is hard law, which is binding, and no deviation is allowed. The hard law that publicly traded companies in Sweden must comply with is the Swedish Company Act, unless otherwise is stated in the law (SFS 2005:551).

According to the Code, the board of directors is responsible for both the organization and the executive management. However, the boards extensive decision-making authority is mainly limited by the legal provisions that gives the annual general meeting exclusive decision-making authority on certain matters, for example the adoption of the income statement and the balance sheet and the election of auditor and board members (Swedish Corporate Governance Board, 2016). According to the Swedish Companies Act, the annual general meeting should be held within six months after the end of the fiscal year (SFS 2005:551). Furthermore, if the specific directives are conducted on the shareholders meeting, the board of directors are obliged to follow these as long as they do not conflict with the Swedish Companies Act (Swedish Corporate Governance Board, 2016).

18

It is not allowed for the board of directors to consist of fewer than three members and one of these are to be appointed to chairman. The chairman has the responsibility to ensure that the board fulfills its obligations and to lead the board in the right direction. Both the Code and the current OMX Nordic Exchange Stockholm have regulations which stipulates that no more than one of the board of directors are allowed in the executive management team of the company or in any of its subsidiaries. However, several companies have no member of the executive management team on the board, although if this is the case, usually the chief executive officer is appointed at the shareholders meeting (Swedish Corporate Governance Board, 2016).

In order for a company to be included in the Nordic Large Cap segment, it is required to have a market capitalization equivalent to 1 billion Euro or more. The segment is revised every six months and the revision is based on the weighted average share price for the period (Nasdaq Inc., 2012). Apart from the requirement regarding market capitalization, there are three major requirements that a company need to fulfill:

• In order to be listed a company must have at least published annual accounts for the last three years in line with the laws applicable in its home country.

• There is a requirement from Nasdaq that when listed on their Stock Exchange, 25% of the financial instruments must be available to the public.

• The company should either be able to demonstrate that it possesses earnings capacity on a business group level and this should be documented or that the company has sufficient working capital available for its planned business operations from the first day of trading and for at least twelve months forward. (Nasdaq Inc., 2018)

19

4. Methodology and Method

_____________________________________________________________________________________

The purpose of this chapter is to explain the methods used in the study and provide a detailed description of where the data is collected, what data that is collected, how the sampling is conducted, and how the data is analyzed.

______________________________________________________________________

4.1 Data Collection and sampling

Our preliminary sample consists of 94 companies listed on the Swedish Large Cap of the Nasdaq Stock Exchange as of 2018-03-07 (Nasdaq, 2018). Data concerning the board of directors are hand-collected from the companies’ annual reports for the time period 2007-2016. As the study aims to analyze the “glass cliff” theory, a time period is chosen in order to include the financial crisis of 2008 to see whether more females are appointed due to the crisis, hence a deductive approach is used by collecting secondary data and compare it to the theoretical framework. By using secondary data, the study takes an external and objectivistic position when analyzing the data and from this demonstrates the social reality in Sweden concerning the companies listed on Large Cap (Bryman, 2012).

The director-level data are hand-collected because there are no databases available, providing full and accurate information. Publicly listed companies in Sweden are chosen due to the requirement of presenting information necessary for the study (Swedish Corporate Governance Board, 2016). The positions of the directors of the board included in the study varies between the companies due to the information available in their annual reports. Hence, chairman, deputy chairman, board of directors, employee representatives and deputy employee representatives are included depending on the companies’ composition of board members. All the positions within the board affects the decision-making and the financial performance of a company, which is why they are included in the study and previous research (Adams & Ferreira, 2009; Joecks et al., 2012) The collected information concerning the directors are their name, year elected, gender and board size.

Data concerning financial performance are primarily collected from the database Amadeus (2018), a database of comparable financial information for private and public companies in Europe. The Amadeus database collects data from official registers, regulatory bodies, annual

20

reports, private correspondence, websites of the companies and press news (Bureau Van Dijk, 2018). The data for the study are based on Amadeus and annual reports, where the data collected from Amadeus are primarily based on annual reports. Annual reports by publicly traded companies in Sweden must comply with the laws and regulations (Swedish Corporate Governance Board, 2016). Moreover, these companies must have an external auditor who approves the content within the annual reports (Bolagsverket, 2016), which ensures the validity of the gathered data for the study (Bryman, 2012).

By selecting the search strategy in Amadeus with the categories “Main stock exchange: Nasdaq OMX - Stockholm” and “Annual market capitalization (th EUR): Last available year, min= 1,000,000 th EUR” 74 companies listed on the Large Cap Stock Exchange in Sweden are received. Observably, companies are missing in Amadeus from the preliminary sample. In this case or if Amadeus does not provide the profitability ratios for the entire time-period, own calculations are made with the formulas used by Amadeus. These calculations are made with the hand collected data retrieved from the companies’ annual reports. As the profitability ratios are given in percentage and not in absolute numbers in Amadeus, the calculations from hand collected data can be compared, even though another currency is used. The formulas for the four profitability measurements are presented as following (Amadeus, 2018):

• ROE= (Profit/ Loss before tax / Shareholders funds) * 100, where Shareholders funds are given by: Capital + Other shareholders funds (-Minority interest).

• ROA= (Profit/ Loss before tax / Total assets) * 100.

• EBIT -margin (Earnings before Tax and Interest) = (Operating Profit/ Loss (=EBIT) / Operating revenue) * 100.

• Tobin’s Q-test= Market capitalization / Total assets.

In order to ensure the reliability of the already calculated profitability measurements of the database Amadeus, the stability is controlled by using the test-retest method (Bryman, 2012). This is done by taking samples from the already calculated measures and comparing them to the own calculated performance measurements based on the company’s Annual report for the same year, to ensure a small variation. Insignificant variations are received from the comparisons, ensuring the stability.

21

To be able to fully cover the time period a selection of performance measurements on an annual basis is made. Due to the annual basis, the study uses a sample criterion that companies must have been publicly listed for the last three years between 2014-2016. Hence, Tobin’s Q-test is available for analyzing the “glass cliff” phenomenon, because the two years preceding the year of the appointment must be examined to see the development of the financial performance. Principally, the statistical tests in the study are independent of time, as they solely focus on the relationship between financial performance measurements and female directors. Nevertheless, if data concerning the board of directors and accounting-based measures are available the years before being publicly listed, these are also included in the study. The data should be available three years preceding the listing, due to the requirements by Nasdaq (2018). The accounting-based measurements are not influenced by market reactions and can therefore be calculated while still being private. One measurement that affects banks, is however EBIT-margin and is calculated by earnings before interest and tax, where banks receive their largest portion of income from interest (Simpson & Kohers, 2002). Due to this, comparing banks EBIT-margin to other firms from our sample would be misleading and therefore excluded as a financial performance measurement for banks. By using these selection criteria, the final sample consists of 80 companies listed on Large Cap in Sweden in 2018, except for EBIT-margin as the banks are excluded, as seen in Table 1. As separate statistical analyzes are conducted for each performance measurements, the different number of observations will not affect the tests.

Table 1

Sampling of companies

ROE can be misleading if a company's liabilities exceed its assets, resulting in a negative equity that has implications for ROE, in the sense that if the equity is negative and the company makes a loss, ROE shows a false positivity. Further, if a company makes a profit and equity is negative, the ROE will be negative. If profit increases or the negative equity improves, the negative result of ROE increases (Hedija, Fiala, & Kunkova, 2017). Due to this, the study excludes ROE calculations for the years that companies present a negative

22

equity. Regarding EBIT-margin, apart from excluding banks, the study excludes abnormal values retrieved from companies with extreme number of operating profit compared to its income. Hence, the study receives different number of observations for the performance measurements which are seen in Table 2.

Table 2

Summary statistics

As the statistical analyzes do not present information about the yearly development, an additional section in the analysis is constructed. Thus, calculations based on the data are made in order to receive average performance measurements, average number of female and male directors on a yearly basis. In addition, number of different positions and proportion of females and males are further presented from the hand-collected data. Hence, the study is able to strengthen the results from the statistical analyzes.

4.2 Performance measurements

The performance measurements used are ROE, ROA, EBIT-margin and Tobin’s Q-test, which Amadeus calculates in Euros. By using calculations that gives numbers in percentage, an observation of the numbers in relation to the companies’ size is given. Even though, the sample consists of companies with a market capitalization of 1 billion Euros or more (Nasdaq Inc., 2012), there are differences in size of their market capitalization. The Amadeus database gives the opportunity to extract numbers into Excel, which are used to collect the available data necessary for the research.

23

Two types of financial performance measurements are used in earlier research, both accounting-based and market-based measurements. By selecting both, the positive and negative aspects of the measurements are captured and gives the study comparable results to earlier research. The selected accounting-based measurements are ROE, ROA and EBIT-margin. ROE display the extent of how effectively companies manage either their own or shareholders capital of the company (Heikal, Khaddafi, & Ummah, 2014). ROA measures how efficient the company turns its assets into profit (Claessens & Tzioumis, 2006). In order to receive an accurate fair value, ROA is often constructed to measure income before extraordinary items, so that companies will not receive a higher score one year because of these items (Adams & Ferreira, 2009; Haslam, et. al, 2010). EBIT-margin is not used in the previous research included in the theoretical framework, however it is included to broaden the research within this field and compare it to the results received by the other measurements to investigate if they differ. EBIT-margin measures how large the operating profit is in comparison to the operating income of the firm (Claessens & Tzioumis, 2006). It is argued that the exclusion of tax and interests is important when evaluating and comparing companies, since it becomes independent of the capital structure (Lie & Lie, 2002; Granata & Chirico, 2010). In addition, the research includes EBIT-margin in order to not be solely dependent on accounting-based measures calculated by profit before tax. As the previous research have not found different results for their accounting-based measurements, the EBIT-margin may provide evidence for new results.

The market-based measurement used is Tobin’s Q-test, measuring how financial policies and events affect the aggregate demand by switching the valuation of assets to its replacement cost, which in other words mean the ratio of market value of firm to the book value of assets (Tobin, 1969). A Tobin’s Q ratio between 0-1, indicates that the replacement cost of the company’s assets has a larger value compared to the market value of the company and hence is undervalued (Amadeus, 2018), whereas a ratio larger than 1 indicates that the market value of the firm exceeds its recorded assets and is therefore overvalued (Singh, Tabassum, & Darwish, 2018). Tobin’s Q-test is proposed to give several advantages compared to accounting-based measurements, because the calculations do not rely on profits from the financial statements that are subject for accounting techniques and influences from managers (Barley, 2014).

24

4.3 Female directors’ relation to financial performance

Collected data are used to explore the relationship between two variables through statistical methods (Williams, 2007). In line with previous research (Adams & Ferreira, 2009; Haslam et al., 2010; Gregory-Smith et al., 2013; Erhardt et al., 2003), female directors impact on financial performance is analyzed the same year as they are elected. This is due to that directors of the board are elected on the annual general meeting, held within six months after the previous fiscal year (Swedish Corporate Governance Board, 2016). Hence, the female directors have at least six months to make a contribution to the financial performance. As a consequence, the performance measurements during the year of female appointments are not affected by the months preceding the appointment from the annual general meeting. The board sizes of the companies differ, leading to different possibilities for the female directors to make an impact. For example, female directors are more likely to have an impact with a large proportion of females in relation to the board size (Joecks et al., 2012). Therefore, the analysis is conducted by the proportion of female directors instead of absolute numbers. To examine the relationship between the variables a bivariate correlation analysis is conducted, as analyzed in previous research (Erhardt et al., 2003; Joecks et al., 2012; Haslam et al., 2010), in the program SPSS. Due to the fact that the study uses interval variables, a Pearson´s r test is conducted (Bryman, 2012). The test examines the relationship between the proportion of female directors against each financial performance measurement. In line with previous research (Adams & Ferreira, 2009; Conyon & He, 2017; Haslam et al., 2010), the study further conducts regression analyzes, because the correlation analysis only indicates if a relationship between variables exists whereas the regression analysis indicates the dependency relationship as a function (Haslam et al., 2010). More specifically, hierarchical regression analyzes are implemented to determine effects of the proportion of females in the boardroom for the performance measurements, while controlling for board size. Therefore, regression analysis is not conducted for the performance measurements with no significant effect from the correlation analysis (Haslam et al., 2010). The selection of board size as a control variable is supported by the theoretical framework (Adams & Ferreira, 2009; Conyon & He, 2017; Erhardt et al., 2003) and extracted from the hand-collected data of the companies. As the proportion of females is calculated by dividing the number of female directors by the size of the board and hence have an impact on the correlation analysis, it is

25

interesting to use board size as a control variable to isolate its effects on the profitability measurements.

In the hierarchical regression the variables are introduced in blocks and in the first block only the control variable (board size) is added and statistically controlled for. In the second block, the independent variable (proportion of females) is entered to the function and is tested for the remaining variance in the dependent variable (performance measurement), as the effects of the control variable is removed (Erhardt et al., 2003).

4.4 The “Glass cliff” phenomenon

To compare the differences in performance prior to a female and a male appointment, a matched sample is conducted, in line with previous research (Adams et al., 2009; Elsaid & Ursel, 2017; Hennessey et al., 2014; Ryan & Haslam, 2005). The data collected for the test includes year of appointment, company name and name of appointee which are extracted from the company data of the study. In addition, the “Industry Classification Benchmark” code (ICB) is used to divide the companies into industries and collected from Nasdaq (2018). The ICB-codes are divided into main categories classified as “Industry” and subcategories classified as “Supersector”. The selection of matched control samples should, if possible, only change one variable (Lyness & Thompson, 1997), which in this study is gender. Hence, a matched male sample is only compared to the same level of directorship, e.g. chairman to chairman, board of director to board of director and employee representative to employee representative. Although, chairmen are compared to deputy chairmen, if no chairman is found, and employee representatives are compared to deputy employee representatives if no employee representative is found.

Firstly, an investigation within the company who appoints a female director is made, to see if a male is appointed during the same time. This is the most similar match, due to the fact that the appointments of the female and male director are in the same financial situation. However, these matched appointments decrease the possibility to find evidence for the “glass cliff” phenomenon. Secondly, if no male appointee is found within the same company, a male appointee within the same year and “Supersector” is selected from the sample. Thirdly, if no male appointee is found within the “Supersector”, a further search within the

26

“Industry” is conducted during the same year. Finally, in some specific cases a search for a male appointee is made within companies connected to each other in terms of similar operational activities when no matched appointment is found within the “Industry”. If these requirements are not fulfilled for the male appointment, both appointments are excluded, as seen in Table 3.

To evaluate the financial performance preceding the appointment of the female and matched male directors, the four measurements are used. Each performance measurement is calculated by the difference of the year preceding the appointment (n-1) and two years preceding the appointment (n-2), divided by (n-2). For example, concerning ROE, if a female is appointed in 2016, ROE is examined as the difference between 2014 and 2015. A positive difference indicates an upturn in the development of the performance measurement and conversely for a negative value. This ensures that the performance measurements are only analyzed preceding the appointment. Hence, the first year examined for an appointment of a female director is therefore 2009. However, due to how the study is constructed, the performance measurements of 2016 do not have an impact on the “glass cliff”, although the study captures the effects from the financial crisis of 2008.

Table 3

Matched sample statistics

From the 80 listed companies on Large Cap in Sweden, a total of 241 female appointments are found during the time period. Due to the selection criteria for the matched sample, the pairs of the performance measurements differ in number. In addition, the exclusion of banks regarding EBIT-margin are still current. Abnormal values are also excluded concerning calculations where the earliest year in the calculation (n-2) is close to zero and the following

27

year (n-1) have either a vigorous up- or downturn, which leads to extreme percentage increases or decreases. This is done, in order to increase the accuracy of the estimation for standard deviation and standard error mean. Due these selection criteria, different number of observations are received for each performance measurement, as seen in Table 3. To test the significance of the difference in performance measurements between genders, a paired sample t-test is conducted for each of the measurements in SPSS. A t-test is also conducted by previous research to test for significance differences (Adams et al., 2009; Hennessey et al., 2014; Ryan & Haslam, 2005). The test measures the difference of the performance measurements in the paired observations, including the female and the matched male appointees. The difference of the development in performance for the pairs are compiled and then a mean is calculated, which simultaneously is statistically controlled for. A negative consequence of conducting this analysis on an annual basis is that the period between the end of the previous year and the annual general meeting of the current year is not included. Previous researches use a time period based on a few months preceding the appointment and contains more recent data compared to this study (Hennessey et al., 2014; Ryan & Haslam, 2005). Furthermore, directors are most likely offered a board position within the company before the annual general meeting and takes the contemporary financial performance into account when considering the position (Adams et al., 2010). As this information are not announced publicly in Sweden, an assumption that directors are both nominated and elected at the annual general meeting is made, in order to be adapted to the yearly based analysis.

28

5. Empirical results and Analysis

_____________________________________________________________________________________

This chapter presents the empirical findings and analyzes the results from the statistical tests conducted in the study. The presentation of the results and analysis are divided into female directors’ relation to performance and the “glass cliff” phenomenon.

______________________________________________________________________

5.1 Female directors impact on financial performance

The Pearson correlation analysis is conducted to examine the relationship between the selected financial performance measurements and female directors. As seen in Table 4, four correlation analyzes are made, one for each dependent variable against the independent variable. Abnormal values for ROE and EBIT-margin are excluded in order to receive more accurate correlations. As seen in Table 2 for minimum and maximum values, the study rejects values over 100 and under -100. Despite the improvements made, the standard deviation for both of these measurements (ROE= 17.76; EBITm= 24.64) are still higher compared to Tobin’s Q-test (3.26) and ROA (11.03). This is mainly because Tobin’s Q-test and ROA are based on total assets, with a high value for companies with a market capitalization over 1 billion €, in relation to the other variable, leading to lower values of the performance measurements and hence the standard deviation. Due to this, different number of observations are received for the dependent variables.

The results from Table 4, present a positive correlation between proportion of females and each of the performance measurements. However, the correlation coefficients are relatively close zero, which indicates that the strength of the relationship between the variables are weak. All the accounting-based measurements receive significant positive correlations, while the market-based measurement receive no significant correlation (p=0,283). Hence, the result is interpreted as no relationship between the two variables exists. Concerning the accounting-based measurements, the coefficient for ROA and proportion of females (r=0,084) is statistically significant at a 95% level. This indicates a 5% chance that no relationship between the two variables exists in the population. Both ROE (r= 0,100) and EBIT-margin (r=0,111) are statistically significant at a 99% level, indicating a 1% chance that no relationship between the two variables exists in the population.

29

Table 4

Correlation statistics.

When testing the variables with positive results from the correlations in hierarchical regressions, as seen in the ANOVA tables of Appendix A, p-values below 0,05 are received, indicating that when both board size and proportion of women are entered, the models are statistically significant for all accounting-based performance measurements. However, when evaluating the first blocks of the hierarchical regression analysis, where only board size is included, a significant impact is received only for EBIT-Margin (p=0,000), as seen in Table 5. This suggests that board size does not make a unique contribution in explaining the variance of ROE and ROA. For the second blocks, proportion of females has a significant impact at the 95% confidence level, on all the accounting-based measurements (ROE: p=0,004; ROA: p=0,014; EBIT-Margin p=0,000). The results indicate that when controlling for board size, all the accounting-based measurements still have significant impacts on the dependent variables.

30

Table 5

Regression statistics

Based on the received results from the correlation and regression analyzes, H1 is accepted. This is due to a significant positive relationship for all the accounting-based measurements and the proportion of females. The findings are in line with several previous researches (Carter et al., 2003; Conyon & He, 2017; Erhardt et al., 2003; Singh et al., 2001), by using different methods and variables. Simultaneously, the findings contradict several of the previous researches with either a negative or no significant relationship between the two variables (Adams & Ferreira, 2009; Gregory-Smith et al., 2013; Haslam et al., 2010). By examining the previous research with positive findings more closely, the most frequent accounting-based measurement used is ROA (Carter et al., 2003; Conyon & He, 2017; Erhardt et al., 2003). The study expands the number of accounting-based measurements by adding two more, both ROE and EBIT-margin. ROE is used in previous research, but solely